OBJECTIVE. The aim of this study was to investigate the association of dose and timing of prenatal alcohol exposure with early language acquisition.
METHODS. We examined language delay in a randomly selected, population-based sample of Western Australian children born in 1995–1996 whose mothers had agreed to participate in a longitudinal study on health-related behaviors and who had completed the 2-year questionnaire (N = 1739). Information on alcohol consumption was collected at 3 months after birth for four periods; the three months pre-pregnancy and for each trimester separately. Prenatal alcohol exposure was grouped into none, low, moderate-heavy and binge (>5) based on the total quantity consumed per week, quantity consumed per occasion, and frequency of consumption. The communication scale from the Ages & Stages Questionnaire was used to evaluate language delay. Logistic regression analysis was used to generate odds ratios and 95% confidence intervals, adjusted for confounding factors.
RESULTS. There was no association between low levels of alcohol consumption and language delay at any time period, although there was a nonsignificant 30% increase in risk when moderate-to-heavy levels of alcohol were consumed in the third trimester. Children exposed to a binge pattern of maternal alcohol consumption in the second trimester had nonsignificant, three-fold increased odds of language delay, with a similar estimate following third trimester alcohol exposure after controlling for covariates.
CONCLUSIONS. This study did not detect an association between low levels of prenatal alcohol exposure and language delay when compared with women who abstained from alcohol during pregnancy. A nonsignificant threefold increase in the likelihood of language delay was seen in children whose mothers binged during late pregnancy. However, the small numbers of women with a binge-drinking pattern in late pregnancy limited the power of this study; studies analyzing larger numbers of children exposed to binge drinking in late pregnancy are needed.
- prenatal alcohol exposure
- fetal alcohol spectrum disorders
- language development disorders
- child development
- longitudinal study
- prospective study
- RASCALS study
Alcohol is known to be a teratogen and can have a range of deleterious effects on children's cognitive, physical, and behavioral development.1,2 The impact of alcohol on the fetus depends on the pattern and quantity of alcohol consumed, the stage of development of the fetus, and a number of sociobehavioral risk factors, such as low socioeconomic status and smoking.3
The full Fetal Alcohol Syndrome (FAS) phenotype is expressed only in children of mothers who had a history of chronic, heavy alcohol use or frequent heavy intermittent alcohol use. However, prenatal alcohol at lower doses can result in a range of milder, although often clinically significant, outcomes.4 Fetal alcohol spectrum disorders is the umbrella term used to describe the full range of alcohol-related harms.5
Language impairment is reported to be a key feature of neurologic damage in children with FAS and fetal alcohol spectrum disorders.6–11 However, the robustness of these conclusions is limited by substantial methodologic weaknesses in terms of design, control of potential confounders, and power. Only a handful of cohort studies have examined the relationship between prenatal alcohol exposure and delayed language development in preschool-aged children without FAS.12–15 The results of these studies have varied, and a consistent association between prenatal alcohol, expressed as an average in ounces of absolute alcohol (AA) per day, and language delay in children in the first 3 years of life has not been demonstrated.12–15 Green et al14 also examined the relationship between a positive Michigan Alcohol Screening Test (MAST) score (indicating alcohol misuse) at the first antenatal visit and language development in children at 2 years of age and found no association.
Language development is a significant developmental milestone for a child, and language delay is an important early indicator of a child's overall development. If language delay is associated with prenatal alcohol exposure, its presence could act as a key indicator for clinicians to review maternal alcohol intake during pregnancy and to screen the child for the presence of other alcohol-related developmental problems.
On the basis of the extant epidemiologic literature, what remains unclear is whether prenatal alcohol exposure is associated with language delay and, if it is, how dose, frequency, and timing of alcohol exposure contribute to the risk of delayed onset of language at 24 months.
To address these questions, we report estimates of the association between antenatal maternal alcohol consumption and subsequent child language development at 2 years of age as assessed by the Ages & Stages Questionnaire16–18 in a large, well-defined and described population-based cohort of non-Indigenous women in Western Australia.
The details of this study have been described previously.19,20 Briefly, a 10% random sample of all women giving birth in Western Australia between 1995–1997 were invited by letter at 12 weeks' postpartum to participate in the Western Australian Survey of Health (known as the RASCALS study), designed to survey health-related behaviors before and during pregnancy and early infancy. After initial contact and consent procedures, an 80.8% response rate resulted in 4861 completed postal surveys. Mothers whose infants were stillborn (n = 20), died in the first 3 months of life (n = 20), or were given up for adoption (n <5) were excluded. The respondents were representative of all women with live births in that period with the exception of a slightly lower percentage of adolescent mothers and of women delivering infants with a low birth weight.21
A 70% random sample of mothers of singletons was drawn from the 1995–1996 cohort (n = 4007) and invited to participate in a longitudinal study. An extra 100 mothers were selected to ensure adequate proportions of low-income single mothers who did not have a partner present. This resulted in 2837 women selected for longitudinal follow-up; 2224 (78%) agreed to participate when their child was 1-year-old and, of these, 1880 (85% of the year 1 sample) completed questionnaires at the 2-year survey. Sociodemographic characteristics of respondents were similar to those of the original cohort in most areas with the exception that they were significantly more likely to earn more than $A25 000 per year (74.5% vs 70.7%; P < .01), be married (79.3% vs 75.6%; P < .03), and be more highly educated (27.8% vs 23.3%; P < .001).22
Children from non–English-speaking households, or households in which languages other than English were spoken (n = 116), and children with severe disabilities and syndromal conditions known to be on the causal pathway of language delay or disorder (n = 25) were excluded leaving a sample of 1739. None of the children in the cohort had received a diagnosis of FAS.
The postal survey questionnaires that were completed at 12 weeks' postpartum allowed for information about maternal alcohol consumption to be collected for the 3-month period prepregnancy and for each trimester separately. For each period, women were asked whether they had ever consumed alcohol and, if yes, how often they drank alcohol (≥5, 3–4, or 1–2 days/week; 1–2 days/month; less than once per month; and never), and the quantity consumed (eg, number of cans, glasses, bottles) on a typical occasion for each of 4 types of alcoholic beverages (beer, wine/champagne, spirits/liqueurs, and fortified wines). Consumption frequency calculations used the lower of the days marked; for example, 3 to 4 days/week was included as 3 days/week. Where respondents used a tick mark instead of indicating a numeric value, a minimum level for the type of beverage, time period, and frequency was applied. As per Colvin et al,20 where a range was given for the number of drinks per occasion, the higher range was used because of the tendency for people to underreport their consumption.
Self-reported maternal alcohol consumption was examined for the 3 months prepregnancy and for each trimester individually. Based on self-reported data, 1 standard drink was considered to be equivalent to 10 g of alcohol. Alcohol consumption was categorized taking into account the total quantity consumed, quantity per occasion, and frequency. Categories were coded as none, low (≤20 g of alcohol per occasion, with a frequency of less than weekly up to 6 days/week), moderate (10 g to <50 g per occasion, with a frequency ranging from less than weekly up to daily consumption), heavy (>20 g to >50 g per occasion, with a frequency ranging from 1 day/week to daily consumption) and binge (>50 g per occasion with a frequency ranging from less than weekly up to 2 days/week (Table 1). Details for the mean, median, and minimum and maximum alcohol consumed per week are described in Table 1. Preliminary analyses indicated that the risk from moderate and heavy prenatal alcohol exposure was equivalent, so these 2 groups were combined because of the small number of heavy drinkers in late pregnancy. The “low” category was coded in line with the recommendation to women who are pregnant or might soon become pregnant set out by the Australian National Health and Medical Research Council (2001) alcohol guideline 11.23
The communication scale from the Ages & Stages Questionnaire16 was used to evaluate language development. This scale is part of a parent report instrument used to screen children for developmental delay. The communication scale comprises 6 items that assess emergent expressive and receptive language abilities using 3 response categories (1): “not yet,” (2) “sometimes,” and (3) “yes.” Performance on the communication scale was scaled using an item response theory model and −1.0 SD was chosen as the cutoff for language delay. Previous work by Zubrick et al22 has shown these items to be a robustly valid and reliable measure of early language emergence in 24-month-old children, with acceptable concurrent validity.
Treatment of missing alcohol data was previously described.20 Briefly, alcohol consumption was not collected in the third trimester for some women who delivered early (n = 27), and for each of these cases the second trimester alcohol consumption was assigned. The average amount of missing data for other variables was low and ranged from no missing data for maternal age, and marital status to 3.8% missing data for income. To address the issue of missing data,24 we conducted data imputation with a multiple imputation procedure by using SAS Proc MI (SAS Institute, Inc, Cary, NC). To account for data imputation procedures we undertook logistic regression using SAS 9.1 (Proc Logistic and Proc Mianalyze).25 These procedures combine the results of the analyses of imputations and generate valid statistical inferences by replacing each missing value with a set of plausible values that represent the uncertainty about the right value to impute.26,27
Although our primary predictor variable was maternal alcohol consumption, several covariates were included in the model: maternal factors (maternal age, parity, education, marital status, smoking for each trimester [none, ≤10 per day or social, >10 per day]), illicit drug use (none versus any) and depression, anxiety, and stress as measured by the Depression Anxiety Stress Scale [DASS])28,29 and family factors (antenatal income, presence of partner in household, parenting ability, and family functioning). The DASS consists of 42 items, with each item rated on a 4-point Likert Scale. Items were summed to generate a score for each of the 3 domains (ie, depression, anxiety, and stress). Severity ratings are based on percentile scores rendering classifications of “normal” (0–78), “mild,” (78–87), “moderate,” (87–95), “severe” (95–98), and “extremely severe.” (98–100). Parenting ability was measured by using the parenting scale, which assesses permissive discipline, overreactivity, and verbosity.30 Family functioning was assessed by using the general scale from the McMaster's Family Assessment Device,31 which measures overall family functioning across 6 areas: problem-solving, communication, affective involvement, affective responsiveness, roles. and behavior control. Higher scores for the parenting scale and Family Assessment Device indicate dysfunctional parenting and family dysfunction, respectively. All variables were entered into the model in a single step, with language development as the response variable.
The study numbers are robust for low to moderate-to-heavy levels of antenatal alcohol exposure. We had 95% confidence and 80% power to detect odds ratios (ORs) of 1.6 for low levels of antenatal alcohol and ORs of 1.8 to 2.0 for moderate-to-heavy antenatal alcohol exposure. The small numbers of binge drinkers in late pregnancy limited the power of our study to detect a significant difference in this group compared with abstainers. The study had the power to detect an OR of 4.7; a fourfold increase in numbers of the exposed group would be required to detect an OR of 2.5.
Ethics approval for the conduct of this study was granted by the Princess Margaret Hospital Research Ethics Committee and the Western Australian Department of Health Confidentiality of Health Information Committee.
Maternal Alcohol Classification
The mean quantity of alcohol consumed per week decreased from the prepregnancy period to trimester 1 and from trimester 2 to trimester 3 for each level of alcohol exposure (Table 1). The biggest decrease in quantity of alcohol consumed occurred for women drinking at low levels, decreasing by 40% over the study period, although the decrease in the minimum and maximum quantities consumed was minimal. The mean level of alcohol consumption for both moderate and heavy drinkers decreased steadily across pregnancy by around 33%. Women in the binge-drinking category did not have a consistent reduction in quantity of alcohol consumed across pregnancy. For these women, the initial reduction in mean quantity of 30% in second trimester, from a mean of 59.8 g/week in prepregnancy to 40.6 g/week, was followed by a 17% increase in average weekly consumption in the third trimester. There were a small number of women in the heavy alcohol category during prepregnancy consuming >1000 g/week (n = 2 [1.2%]) and 400 to 999 g/week (n = 12 [6.7%]) (data not shown). A large drop occurred in the maximum quantity of alcohol reported by heavy drinkers from 2450 g/week during prepregnancy to 345 g/week in first trimester. Binge drinkers consumed around one third of the mean quantity of alcohol of heavy drinkers per week, and the largest drop in maximum consumption was 44%, occurring between the first and second trimester.
Self-reported Maternal Alcohol Consumption
The percentage of women abstaining from alcohol use increased from 15.5% during the prepregnancy period to over 50% during each trimester of pregnancy (Table 2). Although the percentage of women drinking at low levels remained fairly constant from before pregnancy and through each trimester, there were large reductions in women drinking at moderate-to-heavy levels, decreasing from 37.9% prepregnancy to ∼10% in the last 2 trimesters. Similarly, binge drinking reduced from 12.8% prepregnancy to <1% in trimester 3.
Just over half (54.5%) of the mothers were aged ≥30 years of age, 60% were having either their first or second child, 34.8% had postsecondary education, and 79.2% were married (Table 3). Before pregnancy, 27% of women smoked cigarettes, with the percentage decreasing by to 16.8% in the third trimester. Illicit drug use (including marijuana, ecstasy, amphetamines, heroin, methadone, cocaine, LSD, volatile substances) and the use of tranquillizers (both licit and illicit) was reported by 12.5% of women. At the 2-year survey, depression was reported by 9.6% of mothers, anxiety by 5.4%, and stress by 6.8%, although 91.6% of women reported the presence of a partner, and 64.3% had an income less than $A50 000 per annum. Family dysfunction, as measured by the McMaster Family Assessment, was reported in 8.8% of families whereas in 27.8% of families' parenting style was classified as abnormal.
Language Development at 24 Months
Almost 13% of the children had delayed language development as measured by the Ages & Stages Questionnaire. The percentage of children with delayed language was lowest in children of mothers who abstained during the prepregnancy period (12%), those whose mothers consumed low levels during pregnancy (<13%), and for children of women who drank at moderate-to-heavy levels during the prepregnancy and the first trimester (12% and 9.9%, respectively; Table 4). The percentage of children with delayed language was highest for children whose mothers had a binge pattern of alcohol consumption during the second and third trimesters, 29.4% and 33.3%, respectively, and for children whose mothers consumed moderate-to-heavy amounts of alcohol during the third trimester (16.5%).
There was no evidence of an association between low levels of alcohol consumption during pregnancy and an increased risk of having a child with impaired language development compared with women who abstained during this period, after controlling for confounding factors (Table 5). In the prepregnancy period, a nonsignificant 52% increase in likelihood of language impairment was observed with low levels of alcohol exposure. There was no evidence that exposure to moderate-to-heavy levels of alcohol during the prepregnancy period or during either the first or second trimester was associated with an increased likelihood of language impairment. However, a nonsignificant 50% increase in the likelihood of a child having language impairment was detected when this level of exposure occurred during the third trimester (OR: 1.50 [95% confidence interval (CI): 0.90–2.49]).
Binge drinking during either the prepregnancy period or first trimester was associated with a small, nonsignificant increase in the likelihood of language impairment (OR: 1.15 [95% CI: 0.59–2.27] and 1.49 [95% CI: 0.60–3.73], respectively; Table 5). When exposure to binge drinking occurred during the second or third trimester, the risk estimate of language impairment trebled. After controlling for confounding factors, the adjusted OR was 3.00 (95% CI: 0.90–9.93) in the second trimester, with similar results for the third trimester.
Covariates associated with increased risk in the fully adjusted model included parenting in each model and maternal stress (mild) in first and second trimester. There was a nonsignificant trend for maternal age <25 years to be associated with increased risk during pregnancy. No significant interaction effects were observed between the alcohol variable and each of the confounding factors in any time period.
This is the first study to our knowledge to show the importance of the timing, pattern, and frequency of maternal alcohol consumption on the risk of language delay in 2-year old children without a diagnosis of FAS. We did not detect an association between low levels of prenatal alcohol and language delay at any time period when children exposed to low levels were compared with children whose mothers abstained from alcohol during pregnancy. We did, however, demonstrate a nonsignificant dose-response relationship, with a threefold increased risk of language impairment in 2-year-old children who were exposed to binge drinking during either the second or third trimesters, and a 50% increase in risk in the children of women who consumed moderate-to-heavy levels of alcohol during the third trimester after controlling for covariates.
Although our study has a relatively large sample size, two- to eightfold greater than previous cohort studies, with sufficient power to investigate the impact of low levels of alcohol, the numbers of women consuming a moderate-to-heavy or binge-pattern of drinking as pregnancy progressed were small. For example, of the 17 women who were binge drinkers in the second trimester, 10 women continued binge drinking in the third trimester, so we are unable to determine whether the association is because of exposure only in the second or only in the third trimester or in a combination of both trimesters. The wide CIs in the moderate-to-heavy results for third trimester and in the binge results for second and third trimester are a clear indication of the lack of power for assessing these levels of prenatal alcohol consumption on language delay and account for the lack of significance. As such, we cannot rule out the possibility that our finding was because of chance.
Although the findings of this study are in agreement with those of Greene et al,14 who reported no association between language development and low levels of antenatal alcohol consumption, our results differ from their finding of no association with alcohol misuse. This difference may be because of the averaging of alcohol consumption, the use of a positive MAST score to identify alcohol dependence and abuse, and the timing of assessment of alcohol use in their study. Our finding that maternal alcohol consumption reduces as pregnancy progresses is in agreement with other studies32–34 and suggests that using a positive MAST Score at the first antenatal visit may fail to discern an association between binge drinking in late pregnancy and language delay. The MAST will help to identify some of the women whose infants are at higher risk; however, because it was designed to identify alcohol dependence and abuse, it may not be sensitive enough to identify women who have a binge pattern of drinking. Our finding that the average quantity of alcohol consumed by binge drinkers did not reduce until second trimester and then increased in third trimester indicates this is an important group for interventions targeting prenatal alcohol consumption.
The averaging of maternal alcohol consumption during pregnancy to give ounces of absolute alcohol (oz AA) per day but ignoring the pattern and timing of alcohol consumption resulted in the mean level of antenatal alcohol for analysis in the Greene et al study being 0.24 oz AA/d in the first trimester decreasing to 0.18 oz AA/d in the third trimester; an average of half to less than one third of a standard drink per day. Using an averaged estimate of alcohol exposure may conceal the increased risk from moderate-to-heavy and binge levels of alcohol during late pregnancy that we found in our study.
Our study has a number of strengths that overcome many of the limitations of previous studies that have investigated the impact of low to moderate-to-heavy alcohol exposure during pregnancy and fetal harm.35 Importantly, the cohort used for this study is a randomly selected population-based cohort allowing our results to be generalized to the wider population. Other particular strengths include the ability to investigate the pattern of alcohol consumption separately for the prepregnancy period and for each trimester, the high rate of follow-up, the low percentage of missing data, and the collection of a wide range of variables, which allowed us to control for known confounding factors.
Collection of exposure data were by a self-administered questionnaire, which has been shown to elicit more valid responses on socially sensitive issues and reveal more unwanted behavior than interviews.36 There had not been any public health campaign focusing on alcohol and pregnancy either before or during the study in Australia, so respondents are less likely than women in other countries to have viewed this behavior negatively or to have underreported their alcohol consumption because of embarrassment or a fear of being stigmatized. Also, the study collected a range of antenatal maternal behaviors and family factors, so the focus was not on alcohol consumption. Furthermore, the data were collected before the child's language emergence and hence there was no opportunity for the outcome to have influenced recall.
Obtaining an accurate estimate of prenatal alcohol consumption is critically important. Reporting of prenatal alcohol consumption is influenced by the method and the timing of the questions.14,34,36–39 Our data were collected by self-administered questionnaire 3 months after delivery. Although there are reports that higher estimates of alcohol consumption are obtained with postnatal assessment,14,34,37,40 it has been reported that antenatal alcohol interviews provide the most valid estimate to use when investigating neurobehavioral outcome.37 This would suggest that retrospective data collection may have limited our ability to detect an association at lower levels of alcohol exposure. Ideally, this study should be repeated using prospectively collected data on alcohol consumption and with larger numbers.
Self-administered questionnaires have been reported to underestimate prenatal binge drinking.38 Estimation of binge drinking in our survey, however, was not obtained through a specific question on binge drinking but instead calculated from responses on frequency, quantity consumed, type of beverage, and measure (eg, cans, glasses, etc). We are unable to determine from our study whether this method produces a more or less reliable estimate, but if binge drinking was underestimated in our study, the bias would be toward the null.
Although we used a measure of language delay that came from maternal report rather than direct assessment of the child, the communication scale measure from the Ages & Stages Questionnaire is a sensitive and early measure of late language emergence in 2-year-old infants.22 Vocabulary and early syntax are the language properties that are accessible to measure at this age and they provide the earliest markers for risk of language development problems.41 Parent report of a child's current language abilities is the method of choice for assessing emergent language abilities in children at 24 months and is widely used in clinical as well as research contexts. In addition, parent report of language abilities provides a largely untapped opportunity to study this important developmental milestone in large-scale samples.
The numbers of children whose language delay can be prevented through reduction in alcohol misuse during third trimester is relatively small; however, this group of children is likely to be at risk of other alcohol-related problems, of which language delay is just one part of the picture. Identification of at-risk pregnancies before delivery allows the opportunity to encourage reduction in alcohol consumption and may facilitate early intervention.
Finally, we removed the data on 25 children with a disability/syndrome known to be on the causal pathway of language delay or disorder to ensure that our estimate was not confounded by preexisting disability. Although 11 of these children had conditions that may be associated with prenatal alcohol exposure, such as intellectual disability and hearing loss, only 1 of these 11 children was exposed to binge drinking in first trimester, and none was exposed to binge drinking in either the second or third trimester. Fewer than 15% were exposed to moderate-to-heavy levels at any stage of pregnancy. We reanalyzed our data with these 25 children included. The estimates of association did not alter.
This study indicates the importance of timing, pattern, and frequency of prenatal alcohol consumption on the risk of language delay in 2-year-old children who do not have a diagnosis of FAS. The results did not demonstrate any association between low levels of prenatal alcohol exposure and language delay. However, a binge pattern of alcohol consumption during either the second or third trimesters was associated with a nonsignificant threefold increase in the likelihood of language delay and moderate-to-heavy alcohol exposure in the third trimester increased the risk by 50% after controlling for socio-environmental covariates. These results indicate the importance of ensuring that women do not misuse alcohol during pregnancy and that they are aware that the risk to the fetus from binge and heavy alcohol consumption continues throughout pregnancy. Studies with prospectively collected data and larger numbers of women consuming high levels of alcohol during late pregnancy, in particular a binge pattern of alcohol consumption, are needed to provide sufficient power to confirm the findings reported here.
We thank Margaret Wood and Peter Cosgrove for maintenance of the databases. We also thank the staff of the WA Data Linkage Unit for access to the WA Data Linkage System and for their assistance in obtaining the data, and the WA Health Data Custodians for access to the core health data sets. We are very grateful to the parents who have participated in the RASCALS study and for the support of the Telethon Institute for Child Health Research; in particular Dr J.J. Kurinczuk for establishing the study and the RASCALS study team.
- Accepted May 23, 2008.
- Address correspondence to Colleen O'Leary, BSc, MPH, Telethon Institute for Child Health Research, PO Box 855, West Perth, WA 6872, Australia. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
Only a handful of cohort studies have examined the relationship between prenatal alcohol exposure and delayed language development in preschool-aged children without FAS. The results of these studies have been conflicting.
What This Study Adds
This is the first study to show the importance of taking into account both the timing and pattern of prenatal alcohol exposure. Maternal binge drinking during late pregnancy was associated with threefold increased odds of language delay in young children.
- ↵Stratton K, Howe C, Bataglia F. Fetal Alcohol Syndrome: Diagnosis, Epidemiology, Prevention, and Treatment. Washington, DC: National Academy Press; 1996
- ↵Bertrand J, Floyd R, Weber M, et al. Fetal Alcohol Syndrome: Guidelines for Referral and Diagnosis. Atlanta, GA: Centre for Disease Control and Prevention; 2004
- ↵Chudley A, Conry J, Cook J, Loock C, Rosales T, LeBlanc N. Fetal alcohol spectrum disorder: Canadian guidelines for diagnosis. CMAJ.2005;172 (5 suppl):S1– S21.
- Church MW, Gerkin KP. Hearing disorders in children with fetal alcohol syndrome: findings from case reports. Pediatrics.1988;82 (2):147– 154
- Adnams CM, Kodituwakku PW, Hay A, Molteno CD, Viljoen D, May PA. Patterns of cognitive-motor development in children with fetal alcohol syndrome from a community in South Africa [published correction appears in Alcohol Clin Exp Res. 2001;25(8):1187]. Alcohol Clin Exp Res.2001;25 (4):557– 562
- ↵Bricker D, Squires J. Ages & Stages Questionnaires: A Parent-Completed, Child-Monitoring System. 2nd ed. Baltimore, MD: Paul H. Brookes; 1999
- Squires J, Bricker D, Potter L. Revision of a parent-completed developmental screening tool: Ages & Stages Questionnaires. J Pediatr Psychol.1997;22 (3):313– 328
- ↵Squires J, Potter L, Bricker D. The ASQ User's Guide. 2nd ed. Baltimore, MD: Paul H. Brookes; 1999
- ↵National Health and Medical Research Council. Australian Alcohol Guidelines: Health risks and Benefits. Canberra, Australia: National Health and Medical Research Council; 2001
- ↵SAS Institute, Inc. SAS/STAT 9.1 User's Guide. Cary, NC: SAS Institute, Inc; 2004
- ↵Rubin D. Inference and missing data. Biometrika.1976;63 :581– 592
- ↵Rubin D. Multiple Imputation for Nonresponse in Surveys. New York, NY: Wiley; 1987
- ↵Lovibond SH, Lovibond PF. Manual for the Depression Anxiety Stress Scales. 2nd ed. Sydney, Australia: Psychology Foundation Monograph; 1995
- ↵Miller I, Epstein N, Bishop D, Keitner G. The McMaster family assessment device:reliability and validity. J Marital Fam Ther. 1985;11(4):345– 356
- Holzman C, Paneth N, Little R, Pinto-Martin J. Perinatal brain injury in premature infants born to mothers using alcohol in pregnancy. Neonatal Brain Hemorrhage Study Team. Pediatrics.1995;95 (1):66– 73
- ↵da Costa Pereira A, Olsen J, Ogston S. Variability of self reported measures of alcohol consumption: implications for the association between drinking in pregnancy and birth weight. J Epidemiol Commun Health.1993;47 (4):326– 330
- ↵Jacobson SW, Chiodo LM, Sokol RJ, Jacobson JL. Validity of maternal report of prenatal alcohol, cocaine, and smoking in relation to neurobehavioral outcome. Pediatrics.2002;109 (5):815– 825
- ↵Kesmodel U. Binge drinking during pregnancy: is it possible to obtain valid information on a weekly basis? Am J Epidemiol.2004;159 (8):803– 808
- ↵Kesmodel U, Olsen SF. Self reported alcohol intake in pregnancy: comparison between four methods. J Epidemiol Commun Health.2001;55 (10):738– 745
- ↵Verkerk P. Differential misclassification of alcohol and cigarette consumption by pregnancy outcome. Int J Epidemiol.1994;23 (6):1218– 1225
- Copyright © 2009 by the American Academy of Pediatrics